Sex allocation theory predicts that parents should bias offspring sex according to the costs and benefits associated with producing either sex in a given context. Accurately interpreting sex-ratio biases, therefore, requires a precise identification of these selective pressures. However, such information is generally lacking. This may partly explain the inconsistency in reported sex allocation patterns, especially in vertebrates. We present data from a long-term feeding experiment in black-legged kittiwakes (Rissa tridactyla) that allowed us to increase investment capacity for some breeding pairs. Previous findings showed that these pairs then overproduced sons compared with control parents. Here, our aim was to test the underlying assumptions of the 2 appropriate sex allocation models for our context: the “cost of reproduction hypothesis” and the “Trivers–Willard hypothesis.” The former assumes a sex difference in rearing costs, whereas the latter assumes a difference in fitness returns. 1) Independent of feeding treatment, rearing sons was energetically more demanding for parents (as revealed by higher energy expenditure and higher baseline corticosterone levels) than rearing daughters, thereby corroborating the underlying assumption of the “cost of reproduction hypothesis.” 2) Evidence supporting the assumptions of the “Trivers–Willard hypothesis” was less convincing. Overall, our results suggest that drivers of parental sex allocation decisions are probably more related to offspring sex-specific energetic costs than to their future reproductive success in our study species. Assessing the adaptive value of sex-ratio biases requires precise investigation of the assumptions underlying theoretical models, particularly as long as the mechanisms involved in sex-ratio manipulation remain largely unknown.